September 1, 2016 (Vol. 36, No. 15)

We’re Used to Tracking Economic and Political Trends, but Now Audits and Polls Are Getting Proteomic

Protein populations rise and fall from tissue to tissue and from cell to cell as organisms develop and do what they must to maintain health—or regain it, should they suffer illness—and each fluctuation, alone or in the context of other fluctuations, may serve as a signal, an indicator of an underlying dysfunction, a sign that an intervention is warranted, or evidence that an intervention is already having an effect.

Taking the measure of protein populations is the essence of protein profiling, a venerable practice that is growing in sophistication as it becomes more proteomic in character. In fact, the phrase “protein profiling” is used synonymously with “proteome profiling.” Arguably, though, protein profiling is the more general term, since a protein profile that encompasses a subset of the proteome may suffice for a particular purpose.

Besides generality, “protein profiling” has the advantage of modesty. It may be taken as an admission that even the most extensive protein profiles are more sensitive to some proteins than others. In any case, some corners of the proteome remain dark.

While protein profiling may lack proteomic omniscience, it is demonstrating a proteomic outlook. It is part of a shift from protein identification to protein quantification, the measurement of protein expression levels. Moreover, protein profiling now encompasses protein structure-function relationships and protein-protein interactions. Protein profiles may reflect not only the fluctuations in protein levels, but also the shifts between active and inactive protein forms.

As always, protein profiles are about comparisons—diseased vs. healthy, treated vs. untreated, experimental vs. control—but now the comparisons have more depth. They are like census-based analyses that substantiate increasingly fine demographic distinctions. Protein profiling is becoming more refined through the use of advanced experimental and analytical approaches, several of which are summarized here.

Reconstructing Differentiation Trajectories

 “A key limitation at the moment is that we are constrained by the number of proteins that we can measure by flow cytometry in a single cell,” says Berthold Göttgens, Ph.D., professor of molecular hematology at the University of Cambridge. Current flow cytometry approaches allow only up to 20 to 30 protein species to be simultaneously profiled in a cell, and the technology typically provides a snapshot of cellular information.

Dr. Göttgens’ laboratory uses both experimental and computational approaches to study the transcriptional control of normal and leukemic blood stem/progenitor cells. Transcriptional control, Dr. Göttgens reasons, needs to be assessed at the level of single cells, not the bulk of cells. Fortunately, cells of different types—which are, presumably, transcriptionally distinct—may be isolated from the bulk according to their surface

maker proteins. In stem cell biology, population-level profiling provides only population averages and is uninformative about the biology of single cells. The importance of capturing single-cell processes, increasingly recognized in many areas of life sciences, is particularly critical for stem cell research, where progression between distinct cellular states is fundamental for understanding the biology and for developing therapeutic interventions.

In a recent study stemming from practical challenges caused by molecular and functional heterogeneities of murine hematopoietic stem cells, Dr. Göttgens and colleagues combined single-cell gene-expression analyses, flow cytometric sorting, and functional assays to better understand the gene-expression program at the single-cell level. “Establishing a connection between surface marker protein profiles and a whole transcriptome helps use the surface markers to purify viable cells,” explains Dr. Göttgens. “Ultimately, this approach can be used for applications with potential therapeutic benefits.”

By taking advantage of recent molecular-profiling technologies, Dr. Göttgens and colleagues interrogated early hematopoietic stem cell differentiation at the single-cell level in mouse hematopoietic progenitor stem cells. Surface-based cell sorting was used to retrospectively assign cells to one of twelve different phenotypes.

“From these same cells, we also recorded mRNA expression for the entire transcriptome,” details Dr. Göttgens. This helped link single-cell gene-expression profiles with single-cell function. An online repository that incorporated the data provided a resource to visualize lineage-specific transcriptional programs and helped generate an atlas of the early hematopoietic stem cell differentiation at the single-cell resolution.

“Linking molecular profiles with surface marker profiles enabled us to detect protein expression levels without the need to lyse the cells,” explains Dr. Göttgens. Measuring surface protein markers and gene expression in the same single cells, and connecting the two, allowed investigators to use the surface markers to specifically purify distinct populations of cells.

It is relatively easy to generate raw data with the new experimental approaches. The new approaches, however, require laboratories to deal with new data types. “There are no established methods on how to deal with the new data types,” warns Dr. Göttgens. “The next couple of years will see a bottleneck on the computational side.”

Low-Abundance Proteins in Plasma

 “Undoubtedly, the biggest challenge in comprehensive or global protein profiling is the dynamic range,” says Hamid Mirzaei, Ph.D., assistant professor of biochemistry at the University of Texas Southwest Medical Center. For the cellular proteome, the dynamic range fluctuates from one copy to about ten million copies per cell, and it also varies with the experimental system, with major differences being seen between cell lines and biological fluids such as plasma.

Major efforts in Dr. Mirzaei’s laboratory are focused on drug discovery, and these experiments, mostly conducted in cell lysates as experimental models, are seeking to understand the binding sites of specific drugs and their interactions with proteins. Other efforts in Dr. Mirzaei’s laboratory are focused on biomarkers. The majority of Dr. Mirzaei’s biomarker work is performed in plasma.

The difference between the dynamic range of plasma and that of cell lysates is estimated to be about five orders of magnitude. “This is one of the factors that makes it so difficult to profile proteins in the plasma,” explains Dr. Mirzaei.

For example, a peptide that originates from a highly abundant protein may mask peptides that come from a low-abundance protein, and estimates that the 22 most abundant proteins represent about 99% of the plasma proteins illustrate the challenges posed by the dynamic range. Most proteins that are expressed in different tissues reach and reside in the plasma, where their measurement provides prognostic or diagnostic value.

“Even though the plasma proteome is the ultimate proteome,” notes Dr.  Mirzaei, “we are still far from being able to dig really deeply into plasma-type proteomes.”

Another key factor in protein profiling is ionization efficiency, and this is a variable that to an extent is determined by the mass spectrometry equipment. “In terms of biomarkers, the field of mass spectrometry has seen significant improvements over the past ten years,” observes Dr. Mirzaei. “Major companies are providing increasingly accurate and sophisticated mass spectrometers, and this is an aspect that can continuously improve.”

Protein profiling is critically positioned for biomarker identification and development. “Post-translational modifications of proteins are potential biomarkers, as opposed to a proteins themselves,” asserts Dr. Mirzai. In recent years, there has been increasing interest in profiling post-translational modifications, and efforts are underway to significantly improve the ability to visualize lower abundance proteins that carry those post-translational modifications.

Malvern Instruments’ MicroCal PEAQ-ITC was developed for drug discovery applications such as hit validation, lead optimization, and assay development.

Damage-Related Signaling

Investigations led by Stephen J. Elledge, Ph.D., professor of genetics at Harvard Medical School, interrogate phosphorylation events in the budding yeast in response to the DNA damage response. “Phosphorylation,” Dr. Elledge points out, “is relatively easier to profile compared to other post-translational modifications.”

In one recent study, Dr. Elledge and colleagues used quantitative mass spectrometry to capture the substrates and the signaling pathways involved in the cellular response. The scientists generated an unbiased database of phosphorylation targets that contained 133 substrates and was enriched for proteins involved in DNA repair, DNA replication, translation, and cell cycle control.

This work helped uncover the central position that the DNA damage response occupies in the vast cellular signal transduction network and in cellular metabolism. In addition, the work revealed links to TOR signaling, inositol phosphate metabolism, and translational regulation.

“The limiting factors in capturing phosphorylation events will always be sensitivity and the quality of the calls,” advises Dr. Elledge. “And these factors depend on how much material can be obtained.” Obtaining sufficient material for mass spectrometry is particularly critical when profiling post-translational modifications.

For certain post-translational modifications, such as phosphorylation, profiling has been facilitated by the availability of affinity reagents. “Affinity reagents allow investigators to profile the proteome at a much deeper level,” explains Dr. Elledge.

In a study that profiled ubiquitination in response to the DNA damage response using quantitative proteomics, Dr. Elledge and colleagues reported that the replication protein A (RPA) complex is ubiquitinated at multiple sites. The RPA complex is a protein scaffold that binds single-stranded DNA generated during replication fork stalling and facilitates repair. In mammalian cells, RPA complex ubiquitination is required for homologous recombination at stalled replication forks.

The budding yeast and the baker’s yeast, two experimental systems extensively used in Dr. Elledge’s lab, provide ideal models to understand the functional relevance of the proteins of interest. These models take advantage of the availability of knockout strain collections that help interrogate the presence of a protein in a specific cellular pathway.

“The ability to more sensitively detect peptides is critical,” says Dr. Elledge. “One thing that is missing in mass spectrometry analyses is that this technique can identify only those peptides that already exist in the database.” As a result, a vast amount of experimental data is discarded during mass spectrometry experiments.

“This is the dark matter of the proteome,” relates Dr. Elledge. “Figuring out what the dark matter contains will make it a lot easier to make new discoveries on the regulation and modification of peptides.”

One of the key questions is whether there is a way to try to figure out what is being discarded from the vast amount of proteomic data. “It is all there, and we are seeing it,” Dr. Elledge continues. “But we don’t know what we are seeing because we are blind to its significance.”

A multi-protease strategy on the HeLa proteome to improve protein sequence coverage, and to target regions of proteins that do not generate useful tryptic peptides.

Combining Mass Spec and Thermal Shift Profiling

Recent studies have reported that brusatol, a natural compound, can potently and selectively inhibit Nrf2 activity and sensitize several cancer cell types to chemotherapeutic agents. “We became interested in brusatol after several papers showed that it inhibits the Nrf2 transcription factor,” says David Stokoe, Ph.D., senior scientist in discovery oncology at Genentech.

Nrf2, a protein critical for the cellular response to oxidative damage, has dual roles in cancer, inhibiting tumorigenesis in some contexts, and stimulating it in others. These disparate roles are related to mechanisms of activation and dysregulation of the KEAP1-NRF2 pathway. Dysregulation of the KEAP1-NRF2 pathway has been implicated in inflammation and cardiovascular, neurodegenerative, and malignant conditions.

Studies conducted by Dr. Stokoe and colleagues revealed that at submicromolar concentrations, brusatol can powerfully downregulate Nrf2 expression. “Our goal,” recalls Dr. Stokoe, “was to use a new technology, proteomic profiling, to identify the targets of this small molecule.”

Using a mass spectrometry-based strategy, Dr. Stokoe and colleagues interrogated the cellular targets of brusatol in a non-small cell lung cancer cell line. Combining this approach with a cellular thermal shift assay, which exploits changes in the conformation and the thermal stability of proteins when they are bound by small molecules, Dr. Stokoe, together with his Genentech colleagues Kebing Yu, Ph.D., and Don Kirkpatrick, Ph.D., found that brusatol regulates Nrf2 by an indirect mechanism that involves a global inhibition of protein synthesis.

While the expression of most proteins from the proteome decreased, the expression of ribosomal proteins increased, suggesting that brusatol could decrease Nrf2 protein levels by modulating translation. Brusatol decreased the expression of many proteins; notably, proteins with a short half-life were the most powerfully impacted ones. This result validated thermal proteomic profiling by mass spectrometry as a new strategy for target identification and characterization in drug development.

“Although the thermal shift assay could be very powerful, it has a drawback,” admits Dr. Stokoe. “With this approach, a negative piece of information is not informative. If a shift is not seen after a compound is added to a cell, this does not conclusively indicate the compound is not binding.”

Recent years have witnessed the emergence of additional experimental strategies for protein profiling. “There is a lot of excitement about protein activity profiling, where instead of looking at protein abundance one actually interrogates protein activities,” elaborates Dr. Stokoe. Protein activity profiling, which can globally characterize the activity of enzymes in their native environment, relies on the use of a probe that binds the active site of an enzyme, and has been successfully used for enzymes that can be covalently labeled.

“This strategy has been used for phosphatases, deubiquitinases, and other proteins with an available cysteine molecule,” notes Dr. Stokoe. “It will be an emerging area as we move forward.” Another promising strategy for protein profiling is chemical proteomics. With this strategy, a functional group is attached to a small molecule, and then the proteins that it binds to are interrogated.  

Biophysical Riches from Meager Samples

A key determinant in the outcome of protein profiling experiments is the amount of protein that is available for analysis. This determinant depends on multiple variables including the cellular abundance of the protein, the nature of the samples that are to be interrogated, and the capabilities of the experimental approach that is to be used. For example, if mass spectrometry were to be used, ionization efficiency would be a relevant factor.

“When it comes to characterizing proteins, a primary constraint that we observe is that there is often very little sample to work with, especially early in the drug development pipeline,” says Lisa Newey-Keane, Ph.D., life science sector marketing manager at Malvern Instruments. “During early drug candidate screening, the most desirable tests are the ones that provide relevant and useful data using only microliters or even nanoliters of sample.”

The ability to extract information from analytical systems while working with very low sample volumes is not the only challenge in protein profiling. Another challenge is the need to reach high enough sensitivity to deliver accurate data. “Also, if possible, a test should be nondestructive, preserving the sample so that it may be reused for further tests.”

Protein profiling is also challenging because of the dynamic nature of proteins. Proteins have a wide spectrum of half-lives. Also, proteins change in abundance and alter their interactions with binding partners in response to their environment. “Being able to measure proteins under representative conditions, and to know that the tests carried out will produce representative data, is crucial,” explains Dr. Newey-Keane.

Malvern Instruments recently launched a new platform that combines Taylor dispersion analysis and ultraviolet detection for molecular sizing. This approach provides automated protein and peptide stability measurements, has ultra-low volume sample requirements, and is ideally positioned to dynamically characterize proteins, even in complex media.

Many of the proteins that are interrogated in biological fluids present promise as biomarkers or for therapeutic interventions. In translational research, the ultimate goal is the ability to transition from the bench to bedside in a manner that is safe and therapeutically effective. The dynamic characterization of proteins is an instrumental step during this process.

“Delivering a protein to the patient in the form that has been identified as clinically therapeutic requires a detailed consideration of a wide range of factors,” comments Dr. Newey-Keane. “These factors include stability, storage, the impact of temperature, stress, and other processing conditions, as well as the ease of delivery.” 

Previous articleCancer and Unstable Genomes Linked to Junk DNA Changes
Next articleCancer Cachexia Inhibits Cell-Protecting Activity of AMPK Enzyme